| Title |
A Study on the Selection of Representative Models of Subway Stations Using Clustering Analysis Techniques for Energy Performance Evaluation of Subway Stations |
| Authors |
신승권(Seung-Kwon Shin) ; 안승호(Seung-Ho Ahn) |
| DOI |
https://doi.org/10.5370/KIEE.2022.71.10.1427 |
| Keywords |
Urban railway station; Cluster analysis; K-means; Standard model; Machine learning; Energy Performance Evaluation |
| Abstract |
In order to realize the government's plan to increase zero energy early and play a leading role in the low carbonization of railway facilities, Since 2025, the National Railroad Corporation has established a Zero Energy Roadmap for railway stations to promote zero energy building certification. However, currently, standards and tools for evaluating various railway history, including subways, are insufficient in Korea. In particular, railway history has characteristics that are distinct from general buildings, such as the presence of peak loads, continuous erosion, and the possession of safety and functional elements that must be equipped Accordingly, an alternative is needed to overcome the limitations of existing domestic energy performance evaluation tools developed for general building evaluation. Therefore, this paper uses the clustering technique as the first step to develop a decision-making tool that presents the energy performance evaluation and improvement direction of railway history It can represent the characteristics of subway history. In order to select a representative model, data collection and clustering analysis of all 291 subway stations in the Seoul area were performed using Open Data provided by the public data portal. K-means and GMM algorithms were used as clustering analysis methods. As a result of clustering analysis, the K-means algorithm was difficult to use as a result of classification because the data columns were diverse and were not in the form of circular distribution by cluster. Compared to K-means, GMM algorithms can significantly distinguish characteristics such as area, completion year, and platform type the results were shown.
The representative standard stations selected based on the results of GMM clustering were Konkuk University Station (Line 2), Suraksan Station, and Hongdae Station. After that, through simulation of representative models and multiple regression analysis using results, the standard model is developed and used as a tool for initial decision-making such as zero energy construction plan and remodeling plan for subway stations. |